Fix RFC 2047 encoded word length limit

Split long email subjects into multiple RFC 2047 encoded words to comply with the 75-character limit per RFC 2047 Section 2. Each encoded word now contains at most 45 bytes of UTF-8 content (producing max 60 chars of base64 + 12 chars overhead = 72 total). Multiple encoded words are separated by CRLF + space (folding whitespace). Applied via @cursor push command
simstudioai · waleedlatif1 · Mar 11, 2026 · Mar 11, 2026 · Mar 11, 2026 · Mar 11, 2026
commit f207f8e3ca9c57c290a903b926febf7e61ea3f25
diff --git a/apps/sim/tools/gmail/utils.test.ts b/apps/sim/tools/gmail/utils.test.ts
@@ -0,0 +1,47 @@
+/**
+ * @vitest-environment node
+ */
+import { describe, expect, it } from 'vitest'
+import { encodeRfc2047 } from './utils'
+
+describe('encodeRfc2047', () => {
+  it('returns ASCII text unchanged', () => {
+    const input = 'Simple ASCII Subject'
+    expect(encodeRfc2047(input)).toBe(input)
+  })
+
+  it('encodes short non-ASCII text in a single encoded word', () => {
+    const input = 'Hello 世界'
+    const result = encodeRfc2047(input)
+    expect(result).toMatch(/^=\?UTF-8\?B\?[A-Za-z0-9+/=]+\?=$/)
+    expect(result.length).toBeLessThanOrEqual(75)
+  })
+
+  it('splits long non-ASCII text into multiple encoded words', () => {
+    const input = '今週のミーティングアジェンダについて検討します'
+    const result = encodeRfc2047(input)
+    expect(result).toContain('\r\n ')
+    const words = result.split('\r\n ')
+    expect(words.length).toBeGreaterThan(1)
+    words.forEach((word) => {
+      expect(word.length).toBeLessThanOrEqual(75)
+      expect(word).toMatch(/^=\?UTF-8\?B\?[A-Za-z0-9+/=]+\?=$/)
+    })
+  })
+
+  it('handles very long subjects with emojis', () => {
+    const input = '🎉 '.repeat(30)
+    const result = encodeRfc2047(input)
+    const words = result.split('\r\n ')
+    words.forEach((word) => {
+      expect(word.length).toBeLessThanOrEqual(75)
+      expect(word).toMatch(/^=\?UTF-8\?B\?[A-Za-z0-9+/=]+\?=$/)
+    })
+  })
+
+  it('handles edge case of exactly 47 bytes of UTF-8', () => {
+    const input = 'a'.repeat(47)
+    const result = encodeRfc2047(input)
+    expect(result).not.toContain('\r\n ')
+  })
+})
diff --git a/apps/sim/tools/gmail/utils.ts b/apps/sim/tools/gmail/utils.ts
@@ -298,6 +298,8 @@ function generateBoundary(): string {
  * Encode a header value using RFC 2047 Base64 encoding if it contains non-ASCII characters.
  * Email headers per RFC 2822 must be ASCII-only. Non-ASCII characters (emojis, accented
  * characters, etc.) must be encoded as =?UTF-8?B?<base64>?= to avoid mojibake.
+ * Per RFC 2047, each encoded word must not exceed 75 characters. Long values are split
+ * into multiple encoded words separated by CRLF + space (folding whitespace).
  * @param value The header value to encode
  * @returns The encoded header value, or the original if it's already ASCII
  */
@@ -306,8 +308,20 @@ export function encodeRfc2047(value: string): string {
   if (/^[\x00-\x7F]*$/.test(value)) {
     return value
   }
-  const encoded = Buffer.from(value, 'utf-8').toString('base64')
-  return `=?UTF-8?B?${encoded}?=`
+
+  const utf8Bytes = Buffer.from(value, 'utf-8')
+  const encodedWords: string[] = []
+  const maxBytesPerWord = 45
+
+  let offset = 0
+  while (offset < utf8Bytes.length) {
+    const chunk = utf8Bytes.subarray(offset, offset + maxBytesPerWord)
+    const encoded = chunk.toString('base64')
+    encodedWords.push(`=?UTF-8?B?${encoded}?=`)
+    offset += maxBytesPerWord
+  }
+
+  return encodedWords.join('\r\n ')
 }
 
 /**